Enhancing Urban Tree Planting through Air Quality Analysis
Use case scenario
As a: Resident of Melbourne who values sustainable city living and concerned about air pollution and its impact on daily health and wellbeing.
I want to: See key roadside areas in the city selected for tree planting projects based on their potential to reduce airborne pollutants.
So that I can: Breathe cleaner air in urban spaces, benefiting from reduced exposure to traffic-related emissions and a healthier living environment.
By:
Analysing microclimate sensor data to identify areas along Melbourne’s roadways with poor air quality and high exposure to vehicular emissions.
Mapping these pollution-prone zones against existing tree canopy data to locate streets with minimal natural filtration and shade coverage.
Integrating tree planting zone data to pinpoint feasible locations for vegetation expansion that align with municipal greening regulations and infrastructure constraints.
Selecting tree species with high pollutant absorption capabilities and modelling their expected impact on local air quality over time.
Applying geospatial analysis to overlay environmental indicators—such as pollutant levels, canopy gaps, and permissible planting areas—to inform strategic decision-making.
What this use case will teach you
This use case explores how to integrate diverse urban datasets—specifically microclimate sensor readings, tree canopy coverage, and planting zone data—to identify roadside locations where strategic tree planting can significantly improve air quality. By engaging with real-world environmental data, you will address pressing urban health challenges related to pollution exposure.
You will gain practical experience in applying geospatial analysis and leveraging microclimate insights to support data-informed decisions in sustainable urban planning.
The case study focuses on techniques for assessing how vegetation coverage, pollutant concentrations, and permissible planting zones intersect, equipping you with the tools to prioritise greening interventions that deliver measurable air quality benefits.
A key learning outcome will be the development of user-friendly dashboards and interactive visualisations to effectively communicate findings to stakeholders such as urban planners, environmental agencies, and community members.
You will also learn to evaluate the environmental and societal implications of targeted greening projects, enabling you to propose data-driven strategies for creating healthier, more breathable urban environments.
Project Goals and expected outcomes
This project showcases the capability to integrate and analyse multiple open datasets—specifically microclimate sensor data, existing tree canopy coverage, and designated planting zones—to identify roadside locations where new tree plantings can most effectively improve urban air quality.
The analysis will involve spatial and environmental evaluations to prioritise streets and road corridors that currently experience high levels of air pollution and lack sufficient vegetation. These assessments will guide decisions around optimal placement and species selection to maximise air filtration benefits.
The goal is to deliver actionable insights that contribute to cleaner urban air by enhancing tree coverage in strategically selected areas. The project supports Melbourne’s broader objectives for sustainability, public health, and environmental resilience.
A central output will be the development of an interactive, user-focused dashboard that visualises pollutant levels, canopy gaps, and suitable planting zones. This tool will empower urban planners, environmental stakeholders, and policymakers to make data-informed decisions.
Ultimately, the project aims to provide clear, evidence-based recommendations for improving air quality through targeted urban greening—enhancing both the environmental performance of city streets and the everyday wellbeing of Melbourne’s residents.
Data Analysis and Visualisation¶
Tree Canopies 2021 dataset
The below visualisation of illustrates the spatial extent of tree canopy coverage across Melbourne by transforming polygon data into geospatial geometries. The processed shapes were plotted on a static map to highlight zones with dense vegetation.
The chart shows a dense distribution of tree canopies in specific parts of the region, highlighted in green. These visualisations are required for identifying areas that already have substantial green coverage and pinpointing regions that require further intervention to enhance urban greenery. This analysis supports strategic planning for biodiversity conservation, reduction of urban heat islands, and improving the quality of public spaces.
In the below visualisation, segmenting the geographic area into a grid by binning both latitude and longitude values, then counting how many tree canopy points fall into each grid cell. These counts are visualised as a 2D heatmap, where darker green shades represent higher tree canopy density across Melbourne.
The heatmap highlights distinct spatial variations in tree canopy cover across Melbourne. Below are some observations:
- Central and southeast grid cells show the highest canopy densities, indicating robust vegetation corridors in those areas.
- In contrast, the northwest quadrant and outermost cells have very low counts, revealing pockets with sparse or nearly absent tree cover.
These insights provides a view of canopy density in different pockets of the city.
Microclimate Sensor dataset
The figure shows each sensor’s position with a 50 m buffer around it. Next, we will assess how the tree canopy within these 50 m zones influences local pollution levels.
The box-and-whisker format helps identify both the typical range (the box) and the more extreme daily averages (the whiskers and any outliers). The taller boxes or higher medians tend to have more elevated PM2.5 levels or wider day-to-day fluctuations. The spread of the boxes reveals how stable or variable temperatures are on a daily basis, and outliers may point to unusual temperature spikes or dips on specific days.
Distribution of Average PM2.5 per Month Across Devices:
The box plot displays the distribution of average PM2.5 levels across different months for all devices. It highlights the central tendency and variability in pollution levels, clearly showing which months record higher or lower PM2.5 concentrations. Observations from this plot reveal that the months from December to March, likely influenced by the summer season, exhibit a higher median PM2.5. Additionally, the plot shows noticeable outliers, indicating that on certain days, there are unexpected spikes in pollution levels. These variations could be due to transient local events or differences in microclimate conditions across sensor locations.
Building upon the previous analysis, we narrow our focus to January—the month with the highest PM2.5 levels. The dataset is filtered to include only January's records, and a new date column is created to group data by each day. For every device, daily averages for PM2.5 is computed, providing a more granular view of environmental conditions. This detailed approach reveals short-term fluctuations and local effects that might be obscured in broader monthly averages, offering deeper insights for targeted urban and pollution management strategies. Additionally, the analysis aims to observe the variance across all devices to determine whether the differences in daily averages are consistent or if certain locations exhibit significantly different patterns.
Average PM2.5 in January Across Devices:
This box plot clearly indicates that PM2.5 readings vary significantly among devices. Some devices consistently record higher median pollution levels with minimal variability, while others show wider ranges and occasional spikes. This suggests that certain locations are more prone to higher pollution levels, likely due to local factors that contribute to uneven distribution. Overall, the box plot offers detailed insights into daily environmental conditions across devices in January, highlighting the differences that are critical for targeted urban and pollution management strategies.
Tree Planting Zone dataset
The following analysis examines tree planting data through a visual summary that highlights the distribution of segment or zone counts across different project schedules. The data is grouped by project timeframes to provide clear insights into how many segments or zones are at each stage. Custom ordering of the schedule column enhances readability and ensures the visualisation delivers more meaningful insights.
From the chart, we can observe that the tree planting initiatives are at different stages across the zones. Approximately one-quarter of the zones have already completed their tree planting, indicating significant progress in these areas. The dataset reveals that the most intensive planting activity is planned for Years 5-7, which is the peak period for upcoming projects. This is followed by a substantial number of projects scheduled for Years 8-10, reflecting medium-to-long term plans. In contrast, the early-stage schedule (Years 1-4) shows the least activity, suggesting that immediate planting efforts are relatively fewer. Additionally, less than a quarter of the zones have an undetermined timeframe for planting, pointing to some uncertainty or pending decisions in these areas.
The below visualisation combines spatial data—the locations of planting zones—with project schedule details to offer a comprehensive, bird’s-eye view of tree planting across Melbourne. It clearly shows how different zones are progressing, whether they are complete or planned for future periods.
The process begins by verifying that essential location details (longitude and latitude) are available, as these are crucial for accurately plotting the zones on a map. Once confirmed, the data is transformed to align with Melbourne’s mapping system, ensuring each planting zone appears in its correct location. The zones are then colour-coded based on their project schedule, enabling quick visual differentiation of their progress. Finally, a background map of Melbourne is added, providing familiar context and making it easier to understand the spatial distribution and planning of the tree planting projects.
Regressing the values of PM2.5 reduction on Tree Planting Zone Schedule datset using the model trained above
Labeling top 20 zones by PM2.5 reduction out of 639 total zones
Tree Planting Zones with Predicted PM2.5 Reduction Potential
The map visualises Melbourne's scheduled tree planting zones, color-coded by implementation schedule and labeled with their predicted air quality improvement potential. This spatial analysis reveals several important patterns:
Observations
- The zones with highest PM2.5 reduction potential (3.48-3.52 μg/m³) appear clustered in specific neighborhoods, particularly in the central and southern portions of the mapped area
- Different schedule categories appear somewhat clustered, suggesting a neighborhood-by-neighborhood implementation approach rather than dispersed planting
- Segments in the "Years 8-10" schedule frequently show the highest PM2.5 reduction potential, consistent with earlier findings
- The labeled segments (top 20 by PM2.5 reduction) reveal that areas with highest potential improvement have predicted reductions of 3.48-3.52 μg/m³
- Segment 22139 shows the highest predicted PM2.5 reduction at 3.52 μg/m³
Observations and Conclusions¶
Dataset Integration and Analysis¶
This project successfully integrated three complex spatial datasets to understand the relationship between urban tree coverage and air quality in Melbourne:
Tree Canopies (2021): Mapped existing tree coverage across Melbourne, revealing dense vegetation in central and southeast areas with sparse coverage in northwestern quadrants.
Microclimate Sensor Data: Provided hyper-local measurements of PM2.5, temperature, and humidity across 12 sensors, showing significant seasonal variations with highest pollution levels in summer months (December-March).
Tree Planting Zones: Identified 639 planned planting areas across Melbourne with various implementation timeframes (Years 1-4, 5-7, 8-10, and undetermined zones).
Open Meteo Historical Data: Provided reliable baseline measurements of PM2.5, temperature, and relative humidity across Melbourne, enabling calculation of pollution reduction effects. By comparing sensor readings with city-wide values, I quantified the precise impact of tree canopy on local air quality, showing reductions of approximately 3.5 μg/m³ in particulate matter where canopy coverage is substantial.
Key Findings¶
Spatial Patterns¶
Sensors recorded widely varying PM2.5 levels depending on location, suggesting microclimate effects
The existing tree canopy is unevenly distributed, creating opportunities for targeted planting
Areas with higher tree canopy density within 50m of sensors showed measurable reductions in PM2.5 levels
PM2.5 Reduction Potential¶
Our ensemble model predicts approximately 3.5 μg/m³ median PM2.5 reduction from tree planting
Top-performing zones show potential reductions exceeding 4.1 μg/m³
Zones scheduled for Years 8-10 show highest average reduction potential (3.52 μg/m³)
Most beneficial planting locations appear clustered in specific neighborhoods
Implementation Timeline¶
Trees typically begin providing initial PM2.5 reduction within 2-3 years of planting, with measurable significant improvements observed after 5-7 years as canopies develop. This timeline should be factored into planning.
The majority of planting zones are scheduled for mid-term (Years 5-7: 223 zones) and long-term (Years 8-10: 219 zones) implementation
Only 20 zones are scheduled for immediate action (Years 1-4)
177 zones remain without a determined implementation timeline which needs attention
Modeling Approach¶
The blended ensemble of Random Forest and XGBoost models achieved an R² of approximately 0.6, demonstrating that local tree canopy area combined with meteorological variables explains a significant portion of PM2.5 variability, while leaving room for other environmental factors.
Recommendations¶
Implementation Timeline: Consider that newly planted trees require time to mature before delivering full air quality benefits.
Plan Undetermined Zones: Establish concrete implementation schedules for the 177 zones currently lacking defined timelines. These represent nearly 28% of all planting zones and should be evaluated based on their potential PM2.5 reduction impact and integrated into the comprehensive urban forest strategy with clear target dates.
Prioritisation: Focus on zones with highest predicted PM2.5 reduction potential, particularly those in Years 5-10 timeframes
Strategic Planning: Consider accelerating implementation in high-impact areas currently in Years 8-10
Adaptive Management: Continue sensor monitoring and model refinement to optimise outcomes as trees mature and environmental conditions evolve, allowing for data-driven adjustments to planting strategies and maintenance routines based on real-world performance measurements
Further Research: Explore additional variables that might explain the remaining variance in PM2.5 reduction
By implementing these recommendations, Melbourne's urban forest strategy can maximise air quality benefits while creating a more resilient, sustainable urban environment.
References¶
Vaishali, Verma, G. & Das, R.M. (2023) ‘Influence of Temperature and Relative Humidity on PM₂.₅ Concentration over Delhi’, MAPAN, 38, pp. 759–769. doi: 10.1007/s12647-023-00656-8 : https://link.springer.com/article/10.1007/s12647-023-00656-8.
Zhang, H., Wang, Y., Jianlin, H., Ying, Q. & Xiao-Ming, H. (2015) ‘Relationships between meteorological parameters and criteria air pollutants in three megacities in China’, Environmental Research, 140, pp. 242–254 : https://link.springer.com/article/10.1007/s12647-023-00656-8.
City of Melbourne (2021) Tree Canopies 2021 Dataset. Available at: https://data.melbourne.vic.gov.au/explore/dataset/tree-canopies-2021.
City of Melbourne (2023) Tree Planting Zone Dataset. Available at: https://data.melbourne.vic.gov.au/explore/dataset/tree-planting-zone.
City of Melbourne (2023) Microclimate Sensor Data. Available at: https://data.melbourne.vic.gov.au/explore/dataset/microclimate-sensor-readings.
Open-Meteo.com (2023) Historical Weather API. Available at: https://open-meteo.com/en/docs/historical-weather-api.